Soil Erosion under Co#ee Trees with Di#erent Weed

Managements in Humid Tropical Hilly Area of

Lampung, South Sumatra, Indonesia

AFANDI*, Tumiar Katarina MANIK*, Bustomi ROSADI*,

Muhajir UTOMO*, Masateru SENGE**, Tadashi ADACHI***

and Yoko OKI***

* Faculty of Agriculture, University of Lampung, Jl. Sumantri Brojonegoro

No. +Bandar Lampung -/+./, Lampung Province, Indonesia

** Faculty of Agriculture, Gifu University, +�+ Yanagido Gifu /*+�++3-, Japan*** Faculty of Environmental Science and Technology, Okayama University,

-�+�+ Tsushimanaka Okayama 1**�2/-*, Japan

Abstract

Soil loss and surface runo# from farmland under co#ee trees with di#erent weed managements

were investigated in Lampung, South Sumatra, Indonesia during rainy seasons from +330 to +333.

Three treatment practices investigated were as follows : co#ee without cover crop ; co#ee with

Paspalum conjugatum sp. as cover crop, and co#ee with natural weeds. Weed management was

done every two week by clearing all the weeds in co#ee plot, and cutting the weeds around the

co#ee trees with diameter + m for the weedy plots. The results showed that the maximum daily

rainfall and intensity based on +* minutes observation were 2,mm/day and +,*mm/h, however,

only +..,� of rainfall intensities was greater than ,/mm/h (classified as erosive rainfall intensity).A relationship between erosivity index (R) and daily rainfall (X) was found as follows : R�+.0,.(X�+*.3), where R : daily erosivity index (m, · t/ha/h) and X : daily rainfall (mm/day). The runo#ratio for cleanoweeded co#ee ranged from 1.* to +/.3�, and decreased after the second yearbecause of the co#ee canopy growth. The presence of Paspalum conjugatum had reduced runo#

until zero after the third year, whereas in natural weeds plot, runo# became zero after the fourth

year. The highest soil loss was found in cleanoweeded co#ee which reached ,,.1 t/ha in the

second year of experiment. The use of cover crop could suppress soil loss until zero after the third

year in Paspalum plot and after the fourth year in natural weeds plot. However the good

management of weeds as cover crops is necessary due to the bad performance of co#ee growth at

both weedy plots. The average soil loss from cleanoweeded co#ee plot was +.,.mm per year

which was below the soil formation rate in Indonesia.

Key words : co#ee, soil erosion, runo#, erosivity index

+. Introduction

Lampung, located in the southern part of

Sumatra Island, is one of the biggest co#ee

producing districts in Indonesia. About /*� ofco#ee production of Indonesia occurs in

Lampung. From +33* to +33., the average ex-

port of co#ee bean from Lampung was +/3,31*

tons out of -,3,*1* tons of total Indonesian’s

co#ee export (BPD-AEKI, Lampung, +330). In

+333, the co#ee export from Lampung

amounted to +1/,2** tons out of -/.,*** tons

(Bank Indonesia Bandar Lampung, ,***).

The major problem of the co#ee manage-

J. Jpn. Soc. Soil Phys.������No. 3+, p. -�+. �,**,�

� �

ment in Lampung is that the co#ee trees were

mostly planted in steep areas where the soil

conservation is usually not applied very well.

Clearing up all the ground cover under the

co#ee plants is the most popular practice for

weeding management. The farmers perform

such a management practice due to the fact

that the co#ee root system is abundant around

the upper part of the soil surface, so even small

weeds like grass type will influence the co#ee

growth. This reflects on the nutrient compo-

nent of the co#ee leaves due to competition

between co#ees and weeds if the weeds were

left unclear. However, clearing all the weeds

promotes soil erosion since this area possesses

a humid tropical climate, characterized by ab-

undant rainfall and high rainfall intensity con-

centrated on rainy season.

It is well known that cropping systems with

ground cover maintenance, especially in rainy

season, are recommended for farming on slopes

in order to reduce soil erosion. Unfortunately,

the study of soil loss under co#ee plants in the

tropics is limited, and a little information is

available about soil erosion from co#ee areas

in Indonesia. An erosion study by Gintings

(+32,) in Sumber Jaya areas, Lampung, meas-

ured soil loss in Robusta co#ee under natural

condition without any other management. The

co#ee tree used was +0�yr old for 2�monthsmeasurement, and +�yr and -�yr old for 0�month measurement. A 0�month erosionstudy was also conducted in Bali by Pudjiharta

and Pramono (+322) in +/�yr old co#ee trees.The information derived from the studies was

just the amount of soil loss from co#ee planta-

tion in a short period. Some researchers in

Indonesia also have used some grasses as cover

crop, but it was limited for horticultural crop

or annual crops, for example Abujamin et al.

(+32-) used Bahia and Bede grass, and Utomo

(+323) reported the using of King grass in sandy

soil in Indonesia.

Lately, the problem of soil erosion in

Lampung, Indonesia, has accelerated the fol-

lowing political changes. Due to weak law

enforcement during the reformation period in

+332, the public was encouraged to open the

Calliandra reforestation areas. In a very short

period during +333, thousand of hectares of

protected forest in Sumber Jaya were con-

verted into co#ee plantation. The increasing of

co#ee bean price more than three times also

enhanced the forest squattering in +332.

The current controversies on soil erosion in

the co#ee�growing areas in Lampung persistdue to lack of comprehensive research. Since

most co#ee farmers in these areas apply clean

weeding management, a prompt evaluation of

soil loss under existing methods of farming

and other alternatives of weed management

was in need. Thus, the objective of this four

years study was to measure the e#ects of

di#erent weed managements under Arabica

co#ee plants on the extent of surface runo#

and soil loss.

,. Materials and Methods

,. + Study site

This study was conducted during the rainy

seasons from +33/ to +333. The study site was

located at Sumber Jaya District, Lampung

Province, South Sumatra, Indonesia. Geogra-

phically, it is located at +*/�*+’ EL and *.�-.’SL. The slope gradient was around +/�with theelevation of 12*m above the sea level. The

average rainfall recorded from +31. to +332 was

around ,/** mm/year and the air temperature

was around ,,� (Afandi et al., +333).The soil had the initial properties as shown

in Table +. The soil was dominated by clay

fraction in all depths, however the soil bulk

density was very low, which is indicating that

the soil was friable and porous. The color of

soil profile (up to +** cm soil depth) was

dominated by brown to red color which in-

dicated that the internal drainage was very

good. The soil reaction is slightly acid with

moderate cation exchange capacity. The soil

organic carbon was quite high probably due to

the fact that the decomposition rate of organic

matter was relatively slow. Based on the above

������ 3+ �,**,�4

soil properties, the soil could be classified as

Latosol (Indonesian soil classification system)

or Dystrudepts (Soil Survey Sta#, +332).

,. , Treatment plot

The erosion plots were consisted of three

plots with +/�gradient, ,*m slope length and /m width, and bordered by -*�cm height zincmetal sheets. Two collection units were in-

stalled at the lower end of each plot. The first

unit was a ditch with the capacity of *.+m-,

which was connected to the second unit (a tank

and water reservoir) by seven pipes. One pipe,

which was in the center, was connected to a

tank by a siphon and the others overflowed

into the water reservoir box in which a triangle

weir and a water gauge were installed. The

detailed design of the plot was shown in Fig. +.

The treatments were as followed :

(+) Treatment + (cleanoweeded plot) : Ground

surface was always keeping bare by hand

weeding at two weeks interval. This manage-

ment is a general practice in this co#ee planta-

tion area so that this treatment is regarded as a

control.

(,) Treatment , (Paspalum plot) : Co#ee with

Paspalum conjugatum as cover crop. Young

Paspalum conjugatum was transplanted to the

experiment plot in November +33/ and Febru-

ary +330.

Paspalum conjugatum is one of weed species,

gramineous perennial of the South Africa

origin. Its stalks, hard and long, are crawled

over the ground, putting out an irregular roots

from joints. Its growth speed is very fast and

the rhizomes are dense. This plant was widely

used in both private and public park areas in

Table + Initial soil characteristics prior to planting

Depth(cm)

pHH,O

Total-N(�)

Organic-C(�)

CEC(cmol/kg)

Texture (�) Bulk density(g/cm-)Sand Silt Clay

*�+*+*�,*,*�-/-/�0*0*�+**

.43,

.423

.43+

.421

.42/

*4,0

*4+0

*4*3

*4*1

*4*0

-4.2

+420

*423

*42,

*42,

+-4-

343

34-

241

241

,/

,/

,0

,0

,2

,-

+0

+-

+-

+/

/,

/3

0+

0+

/1

*430

*43-

*433

*43-

�

Fig. + The design of experimental plot.

�� : Soil Erosion under Co#ee Trees with Di#erent Weed Managements in Humid Tropical Hilly Area of Lampung, South Sumatra, Indonesia 5

Indonesia.

Instead of these reason, we used Paspalum

because it was abundant in that area and very

easy to manage.

(-) Treatment - (natural weeds plot) : Co#ee

with natural weeds as cover crop.

In each plot, seedlings of Arabica co#ee were

planted with planting distance +./m by ,m on

November +33/. Weed management was done

every two weeks by clearing all the weeds in

cleanoweeded plot, and cutting the weeds

around the co#ee tree with diameter +m for

the weedy plots. Before and after rainy season,

the Paspalum mats and natural weeds were

mowed at +/-cm height. Application of fertiliz-

er and pesticides had been adopted according

to the standard usual practice.

,. - Rainfall erosivity

Rainfall was measured by automatic weath-

er station, which recorded every ten minutes.

We selected +,- rainfall event data from +330 to

+333.

The rainfall energy factor was calculated

using the equation of Wischmeier and Smith

(+312) as follows :

EI�E�I-* �Where E : total rainfall energy in metricoton

meter per hectare (t ·m/ha)

I-* : the maximum -*-min rainfall intensity

(cm/h)

E was calculated as follows :

E�Ek�r �Ek�,+*�23 logI �

Where Ek : kinetic energy in metricoton meter

per hectare per centimeter of rain

(t ·m/ha/cm)

I : rainfall intensity (cm/h)

r : rainfall(cm)

Rain shower less than *./ inch and separated

from the other rain periods by more than 0

hours was omitted from the erosion index,

unless as much as *.,/ inch of rain fell in +/

minutes. According to Wischmeier and Smith

(+312), the value of ,23 (t ·m/ha/cm) was ap-

plied for all intensities greater than 1.0 cm/h, so

equation (-) will be

Ek�,+*�23 log I for I�1.0 cm/h �Ek�,23 for I�1.0 cm/h �Because energy computed in EI equation is

expressed in hundreds of metricoton per ha (t ·

m,/ha/h), the erosivity index (R)becomes :

R�(E�I-*)�+*�, �,. . Soil loss measurement

The sediment in the ditch was pulled out into

a drum with a siphon, thoroughly stirring the

contents in the drum and quickly extracting

/**ml sample with plastic bottle. By measur-

ing the dry weight and the volume of runo#

water stored in the ditch, the soil loss in the

ditch was calculated as follows :

Sd�(A//**)�Vd�+,*** �Sd : total soil loss in the ditch (g), A : dry

weight of /**ml water sample (g), Vd : volume

of runo# water in the ditch (liter), +,*** : con-

version factor from ml to liter

The sediment in the tank was sampled and

the total soil loss in a tank (Sdr) was calculated

using the same procedure as mentioned above.

Total soil loss from erosion plot in an individu-

al measurement was converted into kg/ha as

follows :

Soil loss�(Sd�Sdr�1)�+**�+*�- �+** : conversion factor from +**m*, to +ha,

+*�- : conversion factor from g to kg

,. / Supporting data

Supporting data which also had been collect-

ed in this study were rainfall and co#ee growth

parameters. Rainfall data were collected using

automatic rainfall recorder every ten minutes ;

while co#ee growth parameters (height, diame-

ter of canopy, and co#ee yield) were measured

directly in the field for evaluating the competi-

tion between co#ees and weeds. The type of

weeds and co#ee growth parameter were ob-

served by Sriyani et al. (+333).

,. /. + Weed vegetation

Although the weeds were cut regularly

every two weeks, some weed species were still

appeared in cleanoweeded plot. As presented

in Table , reported by Sriyani et al. (+333), the

dominant weed species in cleanoweeded plot

was Ageratum conyzoides which was an annual

���� 3+� �,**,�6

herb. Instead of Paspalum conjugatum, a few

weed species also occurred in Paspalum plot

including Polygala paniculata, Hedyotis au-

ricularia, and Ageratum conyzoides. Clibadia

surinamense (woody species) whose mean plant

height reached +-0 cm was the dominant spe-

cies (occupy 32� of the total number of weedspecies) in natural weeds plot. The photos of

dominant weed species were shown in Fig. ,.

The weeds started to cover the entire ground

surface (+**� covered) around March +330 atPaspalum plot and on April +330 at natural

weeds plot. After that, the weeds were

managed by spot weeding with diameter +m

around the co#ee tree, and the coverage of

weeds, the ratio of the area occupied by weeds

to total areas, was calculated as 03�. So,during the experiment period, the weeds cov-

ered the soil surface between 03� (after spotweeding) and +**� (before spot weeding).

,. /. , Co#ee growth

The existence of weeds has influenced on the

co#ee growth. The conventional weed man-

agement (cleaning up weed) gave the best per-

formance of co#ee growth as shown in plant

height, canopy diameter and coverage. The

weeds had suppressed the co#ee height at

Paspalum plot about .*� and at natural weedsplot about -*�. The canopy diameter was alsosuppressed as much as /1 � at Paspalum plotand /. � at natural weeds plot.Due to the bad performance of co#ee growth

at both weedy plots, the co#ee yields of the

weedy plots were very low. As reported by

Sriyani et al. (,***), the yield reduction of

co#ee beans in weedy plots in +333 was .*� atPaspalum plot and 1/� at natural weeds plot

Table , Type of weeds species found in each treatment

Clean-weeded plot Paspalum plot Natural weeds plot

Ageratum conyzoides

Borreria repens

Erechtites valeriaifolia

Imperata cylindrica

Oxalis barrelieri

Paspalum conjugatum

Polygala paniculata

Hedyotis auricularia

Ageratum conyzoides

Borreria repens

Clibadia surinamense

Clidemia herta

Chromolaena odorata

Melastoma a$ne

Imperata cylindrica

Fig. , Dominant species of weed in each ex-perimental plot.

�� : Soil Erosion under Co#ee Trees with Di#erent Weed Managements in Humid Tropical Hilly Area of Lampung, South Sumatra, Indonesia 7

compared to cleanoweeded plot.

-. Results and Discussion

-. + Rainfall Pattern

Table - shows that the total rainfall during

four years of experimental period was ./.1.0

mm, with ++-0.3 mm per rainy season. This

rainfall was lower than average rainfall occurr-

ing in this area, i.e. +23- mm per rainy season

(Afandi et al., +333) since El Nino’s occurrence

in +331 caused a long dry season with low

rainfall. In addition, due to plot construction

during the first year, the experiment began

late, in February +33/, whereas the rainy

season usually occurred from October to April

or May.

The daily rainfall distribution (Fig. -)

showed that .2� of the rainfalls occurred lessthan / mm/day or about 0-� was less than +*mm/day. Based on the equation (3) described

later, about -1� of the daily rainfall could becategorized as the daily erosive rain (�+*.3mm/day).

The maximum intensity was ,* mm in +*

minutes or +,* mm/h, which occurred at March

0, +333. According to Hudson (+310), the soil

erosion will be occurred if rainfall intensity�,/ mm/h, and as it was shown in Fig. ., only

+.� of the rainfall intensities which were cal-culated based on +*-minutes time interval

could be classified as erosive (�,/ mm/h).The maximum daily rainfall was 2, mm/day

occurred at January +1, +332, which almost con-

centrated in two hours and its maximum inten-

sity was +-./ mm in +* minutes or 2+ mm/hr.

As shown in Fig. /, this rainfall was fallen in /

Table - Rain erosivity data during . years experiment

Year ofexperiment

R(t · m,/ha/hr)

Rainfall(mm)

Average Rainfallfrom nearest

rainfall station*�(mm)

+st

,nd

-rd

.th

Total

Average

-2/40

/++40

+/.142

3,,4*

--014*

2.+42

./04+

+*1.4,

+10242

+,.24/

./.140

++-043 +23-42

*�Calculated from , rainfall stations (Pajar Bulan and SumberJaya) for rainy season data (Nov4 to May) from +31.�+332

Fig. - Distribution of daily rainfall.

Fig. . Distribution of rainfall intensities basedon +*-minutes time interval of rainfall.

Fig. / Pattern of rainfall Indonesia, occurredat January +1th, +332.

����� 3+� �,**,�8

hours, with intermittent pattern ; twenty

minutes after the rain started, it stopped, and

began again for about two hours and ten

minutes. The intermittent pattern is a typical

rainfall in this area, so several rainfall events

could occur in one day. Due to this rainfall

character, the rain which fell later could be

more erosive although smaller than the previ-

ous rainfall, because the soil has been already

wet and easy to disperse.

-. ,. Erosivity Index

The value of erosivity index ranged between

-2/.0�+/.1.2 and was tabulated in Table -.Since the measurement period for each rainy

season is not the same, the erosivity values

di#er much. In the first year experiment, the

measurement period was only two months due

to the erosion plot construction. A long dry

season due to El-Nino occurred during the

second year of experiment until the end of +331,

thus the amount of rainfall was meager al-

though the length of the rainy season was

normal. Hence, normal value was probably

found in the third and fourth year of measure-

ment, and the values were +,/.1.2 and 3,,.* (ton

·m,/h/ha) respectively.

The rainfall erosivity index as calculated in

equation (0) proved to be the optimal rain pa-

rameter in Indonesia as indicated by Suwardjo

(+32+) and Utomo (+323). Utomo (+323) reported

the values between 3+-�+,0-- (ton·m,/h/ha) forseveral areas in Indonesia which agreed with

the normal values in this experiment.

-. - Relationship between daily rainfall

and erosivity index

Based on +,- rainfall events from +33/ to +333

(Fig 0), a relationship between erosivity index

which was calculated using equation (.), (/),

and (0) and observed daily rainfall was derived

as follows :

R�+.0,. (X�+*.3) �with r,�*.2,Where R : daily erosivity index (t ·m,/h/ha)

X : daily rainfall (mm)

From this equation which shows the linear

relationship between R and X, it can be in-

ferred that the rainfall will be erosive if the

amount of rainfall exceeded +*.3mm per day.

Due to the fact that the rainfall data available

in Indonesia was mostly daily rainfall, some

researchers had proposed rain erosivity index

using daily rainfall data. The rainfall erosion

index widely used in Indonesia as rain

erosivity factor for applying USLE was intro-

duced by Bols (+312) as follows :

EIm�0.++3 (CHb)+.,+ (HH)�*..1 (CH,.)*./- ��where :

EIm : monthly rain erosivity index (t ·m,/h/

ha)

CHb : monthly rainfall data (cm)

HH : monthly rainfall event (days)

CH,. : the amount of maximum rainfall in ,.

hours during one month (cm)

Applying the observed rainfall data to equa-

tion (+*), the values of EIm between /-, and

-,-1* (t ·m,/h/ha) with a mean of +,2./(t ·m,/h/

ha) were obtained in this experiment. These

values were two fold higher than Wischmeier

and Smith’s rain erosivity index (equation (.)

and (/)). As shown in Table -, the Wischmeier

and Smith’s rain erosivity index ranged from

-2/.0 to +/.1.2(t ·m,/h/ha) with a mean of 2.+.2

(t ·m,/h/ha). An estimated rain erosivity index

between 3** and ,,/** (t ·m,/h/ha) were also

found for Lampung areas (Manik and Afandi,

+332) that were also higher than the values

found in this experiment.

-.. Surface runo#

The presence of weeds as well as the diame-

ter of co#ee canopy influenced the runo# ratio

very significantly. As shown in Table ., in the

Fig. 0 Relationship between daily rainfall anderosivity index.

�� : Soil Erosion under Co#ee Trees with Di#erent Weed Managements in Humid Tropical Hilly Area of Lampung, South Sumatra, Indonesia 9

first year of experiment, the average runo#

ratio was very high (+/.3�, +./�, and 3.*� forcleanoweeded plot, Paspalum plot, and natural

weeds plot respectively) compared to the other

years, since the soil surface was relatively

open. The runo# ratio from the second to

fourth year experiment was 3.* �, 3.0 � and1.*� for cleanoweeded plot ; *� for Paspalumplot ; and *.-�, -.3� and *� for natural weedsplot.

Although the total runo# of cleanoweeded

plot increased during the second year of exper-

iment, its runo# ratio decreased due to the

increase of the co#ee tree canopy. The increase

in co#ee coverage also reduced the runo# ratio

in the fourth year of experiment although the

rainfall was fairly high.

The runo# ratio for cleanoweeded plot was

higher than that found by Gintings(+32,) in

Sumber Jaya areas. Without cover crops, he

found a runo# ratio of +.12� for +�years oldco#ee with /3�0-� slope gradient, and -.-3 �for -�years old co#ee with slope gradient 0,�00�.It was interesting to note that the Paspalum

conjugatum and natural weeds could reduce the

surface runo# drastically. Paspalum con-

jugatum is a grass weed type, so the canopy

fully crept over the ground surface and its root

system was very dense and large. As a conse-

quence, Paspalum conjugatum could intercept

the rainfall e#ectively and would act as runo#

barrier, thereby inhibiting the surface runo#

significantly. Since the internal drainage of

soil profile was very good, the runo# water

would infiltrate into the soil.

As shown in Table ,, a weed of tree type

called Clibadia surinamense was present at nat-

ural weeds plot and could behave as a shading

tree. As consequence, the soil surface was not

fully covered with grass weeds type, so the

surface runo# was greater than Paspalum plot.

The e#ectiveness of cover crop in reducing

runo# was clearly shown in Table .. In

Paspalum and natural weeds plot, the increas-

ing of rainfall did not increase runo# height.

However, there exists a tendency in clean�weeded plot that runo# height would increase

with rainfall.

-. / Soil loss

The amount of soil loss in every month and

the total soil loss during the experimental

period are shown in Fig. 1, Fig. 2, and Table /.

In the first year of experiment, the soil loss

from each treatment during the three months

observational period (Feb�April +330) was asfollows : 2.2 t/ha in cleanoweeded plot, ,.1 t/ha

in Paspalum plot, and +./ t/ha in natural weeds

plot. Before the weeds covered the ground

surface, there was a little di#erence in soil loss

as shown in February +330 ; soil loss from

Paspalum plot was higher than the other plots

due to the fact that the ground surface was not

covered yet and was disturbed by transplant-

ing Paspalum seedlings at this plot. One month

later (March +330), this situation was changed

drastically after the weeds began to cover the

ground surface ; the soil erosion from clean-

Table . Runo# and runo# ratio of each rainy season

RainySeason

Period Rainfall(mm)

Runo#(mm)

Runo# Ratio(�)

Clean-weeded

Paspalum Naturalweeds

Clean-weeded

Paspalum Naturalweeds

+st

,nd

-rd

.th

30 Jan.�30 Apr.30 Oct.�31 Apr.32 Jan.�32 Jun.33 Jan.�33 Jun.

.0140

+*124,

+1034.

+,.24/

1.4/

314/

+034-

214-

042

*4/

*4*

*4*

.,4+

-4-

024-

*4*

+/43

34*

340

14*

+4/

*4*

*4*

*4*

34*

*4-

-43

*4*

Total ./0-41 .,240 14- ++-41

������ � 3+ ,**,�10

oweeded plot increased sharply.

The second year experiment began in Octo-

ber +330 and continued until April +331. The

soil loss from co#ee plot with general weed

management (cleanoweeded plot) in the second

year of experiment was ,,.1 t/ha, however it

could be *.*,2 t/ha if covered with Paspalum

conjugatum and *.+0 t/ha in natural weeds plot.

The weeds had already covered totally the

ground surface in Paspalum plot and natural

weeds plot (+**� covered). However therewere big di#erences in weeds species (Table ,),

which a#ected the amount of soil loss.

Paspalum plot was dominated by Paspalum

conjugatum, which was very e#ective in pre-

venting soil erosion due to the fact that ground

surface was covered densely with Paspalum

whose canopy and root system would act as

runo# barrier and catch the direct raindrops

and prevent surface runo# from flowing over

the ground surface. The natural weeds were

dominated by woody species and not so dense

compared to Paspalum plot. Thus, a chance for

rainfall to pass the canopy and overflow on the

ground surface among the weeds existed.

Due to the long dry season in +331 because of

El-Nino e#ects, the third observation began in

+332 for rainy season of +331/+332 and the

fourth observation began in +333 for rainy

season of +332/+333. The soil loss from co#ee

with weeds was very small compared to clean-

oweeded plot. The soil loss in +332 from clean-

oweeded plot was 3.+ t/ha followed by natural

weeds plot *.01 t/ha, and no erosion was found

in Paspalum plot. In +333, the soil loss from

cleanoweeded plot was ..2 t/ha, and there was

no soil loss in both weedy plots.

Compared to the other research data with

di#erent crops in Indonesia, the soil loss from

the co#ee areas was meager. However, com-

paring with the findings of Gintings (+32,) in

co#ee areas in Sumber Jaya, soil loss appeared

higher. The maximum soil loss in this experi-

ment was found in cleanoweeded co#ee as

much as ,,.1 t/ha. Utomo (+323) reported that

the soil loss from a bare plot could be .,- t/ha.

During the four years experimental period

from +310 to+32*, Abujamin et al. (+32-) found

soil loss from bare plot to be +3-./�./, t/ha.

Fig. 1 Monthly soil loss under co#ee tree with di#erent weed management.

Fig. 2 Soil under co#ee tree with di#eremtweeds management.

�� : Soil Erosion under Co#ee Trees with Di#erent Weed Managements in Humid Tropical Hilly Area of Lampung, South Sumatra, Indonesia 11

However, they also reported that the use of

Bahia grass (Paspalum notatum) strip with +m

width could suppress soil erosion to zero in the

second and third year of experiment, whereas

*./ m strip of Bede grass (Brachiara decumbens)

suppressed soil erosion until zero after the

fourth year of experiment.

Gintings (+32,) found that the soil loss from

+�year old Robusta co#ee with /3�0-� slopegradient was +.3. t/ha which was very low

compared to the soil loss from cleanoweeded

co#ee (,,.1 t/ha). On the other hand, the soil

loss from -�years old co#ee with slope gradient0,�00� measured from May to October was+./1 t/ha (Gintings, +32,), which was lower than

the soil loss from cleanoweeded co#ee with

similar age which was 3.+ and ..2 t/ha.

The e#ect of co#ee canopy on reducing soil

loss was demonstrated clearly in Fig. 2 and Fig.

3. Table . showed that in cleanoweeded plot,

runo# was fluctuated but almost constant

from the second to the fourth year of experi-

ment. However, the soil loss tended to decrease

as shown in Fig. 2. It means that although the

runo# ratio was relatively unchanged after the

second year, the sediment which carried by

runo# became smaller year by year as shown

in Fig. 3. Increasing the co#ee canopy was the

major factor that was responsible of these phe-

nomena. The co#ee canopy intercepted the

rainfall, hence the raindrops did not directly

strike the soil surface and soil aggregate did

not disperse, so runo# would carry less sedi-

ment.

-. 0 Soil loss tolerance

As shown in Table /, the soil loss from this

experiment was very low ; the total soil loss

from cleanoweeded plot was around ..30mm

for four years experiment or +.,.mm per year.

This value is below the soil formation rate in

Indonesia, which is estimated about ,mm per

year (Shah, +32,). If we took the average soil

loss only for three years observation (because

the +st experiment started too late), the aver-

age soil loss was around +,.31 t/ha or +.-/mm/

yr. This value was also lower than the soil loss

tolerance for that area which was estimated

between ,./�-.1mm per year (Manik et al.,+331) or ,,.. t/ha per year for Indonesia (Utomo,

+33.).

.. Conclusion

The four years soil erosion research con-

ducted in a hilly humid tropical area of

Lampung, Indonesia, showed that only +..,�of rainfall intensities could be categorized as

the erosive intensity (�,/mm/h). The maxi-mum rainfall intensity was +,*mm/h, and the

maximum daily rainfall was 2,mm. A relation-

ship between rainfall erosivity index (R : t ·m,/

ha/h) and daily rainfall (X : mm) was deter-

mined as follows :

R�+.0,. (X�+*.3)Soil loss from cleanoweeded plot ranged

from ..2 t/ha to ,,.1 t/ha. The total soil loss

from cleanoweeded plot was ..1*mm during

the four years experimental period, and +.+1

Fig. 3 Sedimentation of each plot.

Table / Total soil loss under co#ee withdi#erent weed management

Unit (mm)

Year ofobservation

Clean-weededplot

Paspalumplot

Natural weedsplot

+st

,nd

-rd

.th

*43+

,4-0

*4/*

+4+3

*4,2

*4**

*4**

*4**

*4+0

*4*,

*4**

*4*1

Total .430 *4,2 *4,/

Average(mm/year) +4,. *4*1 *4*0

����� 3+� �,**,12

mm/year was lower than the soil formation

rate in Indonesia.

The use of Paspalum conjugatum and natural

weeds as cover crop under co#ee trees was

very e#ective in controlling soil erosion. The

soil loss from Paspalum plot was ,.1 t/ha

during the first year of experiment and became

zero after the third year. However, the soil loss

from natural weeds plot in the first year was +./

t/ha and zero after four years. Instead of

weeds coverage, the reduction of soil loss after

three years experiment was also caused by

co#ee canopy through its rainfall interception.

The presence of weeds and the increasing of

co#ee canopy as the co#ee growth influenced

the runo# significantly. Paspalum conjugatum

could suppress runo# until zero after the third

year of experiment, and in case of natural

weeds it suppressed until zero after four years.

Runo# in cleanoweeded co#ee decreased year

by year due to co#ee canopy growth, and

runo# ratio became almost constant around 1�+*� after the second year of experiment.Both weedy plots showed the bad perform-

ance of co#ee growth. However, the use of

Paspalum conjugatum as cover crop in co#ee

garden was promising if managed optimally.

Since Paspalum has a shallow roots, it com-

petes with co#ee’s root, thus these species have

to be introduced when the co#ee trees have

gained enough height (about three years old).

Paspalum could be planted along strips be-

tween the co#ee plants or around the co#ee

canopy.

Acknowledgments

Appreciation is extended to Ministry of Edu-

cation, Culture, and Sport of Japan, for funding

this study, and also to Dr. N. Sriyani, Mr. H.

Suprapto, Mr. H. Susanto, and Mr. A.T. Lubis

for supporting the weeds and some co#ee data.

Reference

Abujamin, S., Adi, A. and Kurnia, U. (+32-) : Perma-nent grass strip as one of soil conservationmethods. Soil and Fertilizer Research News,Center for Soil Research, Ministry of Agricul-

ture, Indonesia, + : +0�,*. (in Indonesian).Afandi, Gafur, A., Swibawa, I.G. and Purnomosidhi,

P. (+333) : Baseline biophysical informationabout the Tulang Bawang watershed area,North Lampung. Proceeding of the Manage-ment of Agrobiodiversity in Indonesia for Sus-tainable Land Use and Global EnvironmentBenefits. ASB-Indonesia Report Number 3.Bogor, Indonesia : 10�+3,.

Bank Indonesia Bandar Lampung (,***) : Financialeconomic�monetary statistics for LampungProvince Region : p. ,/. Bank Indonesia Ban-dar Lampung.

BPD-AEKI Lampung (Indonesian Association ofCo#ee Exporter Lampung Branch) (+330) : Ex-perience of marketing and the prospect ofLampung co#ee : pp.+0�-2. Roadshow of Tech-nology of Arabica co#ee development in West-ern Lampung, Liwa, Western Lampung, Janu-ary (in Indonesian).

Bols, P.L. (+312) : The Iso-erod on Map of Java andMadura, Soil Research Institute, Minsitry ofAgriculture Indonesia, Bogor.

Gintings, A.G. (+32,) : Surface run-o# and soil ero-sion under co#ee plantation and natural forestat Sumber Jaya-North Lampung. Forest Re-search Institute, Ministry of Agriculture Indo-nesia, Report No. -33 (in Indonesian).

Hudson, N.W. (+310) : Soil conservation. BatsfordLtd, London.

Manik, K.E.S., Susanto K.S. and Afandi (+331) : Esti-mation of allowable soil loss in two dominantsoil in Lampung. Journal of Tropical Soils II(.) : 33�+*0 (in Indonesian).

Manik, K.E.S. and Afandi. (+332) : Rain erosivityindex in Terbanggi Besar, Central Lampung-Analysis at three long dry seasons. Environ-ment and Development +2(.) : ,2,�,21 (in Indo-nesian).

Pudjiharta, A.G. and Pramono, I.B. (+322) : Runo#and soil erosion under natural forest standsand co#ee plantations at Tabanan, West Bali.Forest Research Bulletin .32 : +�2 (in Indone-sian).

Shah, M.M. (+32,) : Economics Aspects of soil ero-sion and conservation. Technical Note No. ,1.Centre for Soil Research, Bogor, Indonesia.

Soil Survey Sta# (+332) : Keys to soil taxonomy 2th

: p1-. United States Department of Agricul-ture. Washington D.C.

Sriyani, N., Suprapto, H., Susanto, H., Lubis, A.T.and Oki, Y. (+333) : Weeds population dynam-ics in co#ee plantation managed by di#erentsoil conservation techniques. Proc. of Interna-tional Sem. Toward Sustainable Agriculture in

�� : Soil Erosion under Co#ee Trees with Di#erent Weed Managements in Humid Tropical Hilly Area of Lampung, South Sumatra, Indonesia 13

Humid Tropics Facing ,+st Century. BandarLampung, Indonesia, September ,1�,2 : /+-�/,*.

Sriyani, N., Suprapto, H., Susanto, H., Lubis, A.T.and Oki, Y. (,***) : The development of sus-tainable biological production technologiesharmonized with regional environmental con-ditions in East Asia. Final Report of the Great-in-Aid for Creative Basic Research from Minis-try of Education, Culture, and Sport of Japan :+3.�+31.

Suwardjo (+32+) : The role of plant debris for soil

and water conservation in annual farm land.Ph.D. desertation. Bogor Agriculture Universi-ty, Bogor, Indonesia (in Indonesian).

Utomo, W.H. (+323) : Soil conservation in Indone-sia, a record and analysis. Rajawali Press, Ja-karta (in Indonesian).

Utomo, W.H. (+33.) : Erosion and Soil conservation.IKIP Malang, Press Malang (in Indonesian).

Wischmeier, W.H. and Smith, D.D. (+312) : Predict-ing rainfall erosion losses. A guide to soil con-servation planning. USDA Agric. HandbookNo. /-1, Washington D.C.

��������������������������������� !"#$#%&'()*+

������ *���� �� ��� *����� ���� *����� ��� *����� **� !"# ***�$ %& ***

*'�(�)*+*,**-.)*+*,

***/0)*1234*,

, -

�567��8���'9'�(�:; +330?@A +333?9BC9DE> FGHIJKL3M9NOPOQ@A9RSTUVWXYZ[V\]^_I` abcd> �XYe9JK\fghK_INOPOQ> �ijklW_=JKm Paspalum conjugatumnXYe\ij_INOPOQ> �opkq9JKnXYe\ij_INOPOQnrs` JK9L3d ,tE: +u9�vnwHI` abc �ndXYe9JK\fg:hx_> yz9abc �W �ndNOPO{|}9~ +m9}\hK_I` ]^DE:;sXc9)BVd 2,> ) +*EBVvd +,*mm/hnrHI` _@_> gB9 +..,�9)Bv> RSTUq9WGsBv ,/mm/h\uH= NOPO9{¡9¢:W£GH=¤¥_I` JK:¦H=XYe\ij§sWXYZ[d¨_©¤¥_> Paspalumabcnd -?ª9BC«9Z[ *�:GHI9:¬_> opkq9JKabcnd .?ª@AZ[ *�:GHI` RSTUV)WGHI9d ,?ª9BC9fghKcn> ,,.1/hanrHI`ijkl\®¯§sWRSTUd¨_©°±²> Paspalumabcnd -?ª«@A>opkq9JKabcnd .?ª@ARSTUq_G@HI` fghKc9RSTUVdRS³:_=+?E: +.,.mmnrz> �567�9RSq´v\MuH= RSTU> Z[> µU¶

µ·?¸ : ,**+? ++¸ + µ3?¸ : ,**,? - ¸ ,3

RS9l3¶ ¹ 3+º ,**,14